A variety of semiconductor devices, such as, bipolar transistors, rectifying diodes, light emitting diodes (LEDs), lasers, photodetectors, solar cells and others, require the formation of doped semiconductor junctions for their operation. Most semiconductor devices in use today, both inorganic and organic, are in part or completely formed using expensive vacuum deposition processes. There are ongoing efforts to find a low cost manufacturing process, but to date, device performance has been inadequate for market needs. Therefore, there is a need for a low cost technique of forming high quality inorganic doped semiconductor junctions for use in semiconductor devices.
In general, both n-type and p-type materials can be referred to as charge transport materials, and the layers of a device containing such materials can be referred to as charge transport layers. An n-type material typically has an excess of conduction band electrons, and as such is also referred to as an electron transport material. Furthermore, an n-type semiconductor is a semiconductor in which electrical conduction is due chiefly to the movement of electrons. A p-type material typically has an excess of “holes”, and as such is also referred to as a hole transport material. Furthermore, a p-type semiconductor is a semiconductor in which electrical conduction is due chiefly to the movement of positive holes. The doping levels of the charge transport layers are typically set so that they are highest when the layers are in contact with metals (in order to assist in forming ohmic contacts). In the case where the layers are in contact with the anode or cathode, the charge transport layers are also typically called contact layers.
Semiconductor diode devices have been in use since the late 1800s. Most modern diode technologies are based on semiconductor p-n junctions, or contact between p-type and n-type semiconductors. However, many types of electronics would benefit from lower cost doped semiconductor junctions.
In the field of photovoltaic devices, current devices employ thin layers of semiconductor material, e.g., crystalline silicon, gallium arsenide, or the like, incorporating a p-n junction to convert solar energy to direct current. While these devices are useful in certain applications, their efficiency has been somewhat limited, yielding conversion efficiencies, e.g., solar power to electrical power, of typically marginally better than 10-20%. Although efficiencies of these devices have been improving through costly improvements to device structure, the relative inefficiency of these devices, combined with their relatively high cost, have combined to inhibit the widespread adoption of solar electricity in the consumer markets. Instead, such systems have been primarily used where conventionally generated electricity is unavailable, or where costs associated with bringing conventionally generated electricity, to a location where it is needed, more closely match the costs of photovoltaic systems.
Despite the issues with current photovoltaic technology, there is still a desire and a need to expand usage of solar electricity. In particular, there is generally a need for an improved photovoltaic cell that has one or more of: increased energy conversion efficiency, decreased manufacturing costs, greater flexibility or reasonable durability or longevity. In fact, as disclosed in U.S. Pat. No. 7,087,832 Scher et al. disclose the use of coatable nanoparticles in a polymer binder for use in photovoltaic devices. However, the performance of these devices were not reported, and the conductivity of such a mixed photoactive layer is expected to be low due to the high resistivity of the polymeric binder. An example of the performance of devices with these hybrid absorber layers is an efficiency of ˜1.5% under AM 1.5 excitation (J. Liu et al., JACS 126, 6550 (2004)). Recently, an all inorganic solution processed solar cell was formed from CdSe and CdTe quantum rod nanoparticles, but again the efficiency was very low at 3% even after sintering the films at 400° C. for 15 minutes (I. Gur et., Science 310, 462 (2005)). A large part of the low efficiency was undoubtedly caused by the films being insulators (even after sintering) due to the lack of doping. For both CdTe and CuIn1-xGaxSe2-yS (CIGSS) solar cells, the window layer is typically n-CdS (N. G. Dhere et al., J. Vac. Sci. Technol. A23, 1208 (2005)). Both doped and undoped forms of CdS have been used in the devices and a preferred deposition technique has been chemical bath deposition (CBD). Even though a solution processed technique, CBD involves dunking the entire wafer into a bath, which can be acidic or basic, for periods up to hours. In addition, the process is inefficient with respect to usage of its starting materials. In summary for the cited photovoltaic devices, either all or parts of the semiconductor junction were produced by low cost processes. However, either the resulting photovoltaic performance was sub-par or the deposition process had major drawbacks.
FIG. 1 gives a schematic of a typical prior art LED device 105 whose basis is a pin diode. All of the device layers are deposited on a substrate 100. Above the substrate 100 are a p-contact layer 110, a p-transport layer 120, an intrinsic emitter layer 130, a n-transport layer 140, and a n-contact layer 150. An anode 160 makes ohmic contact with the p-contact layer 110, while a cathode 170 makes ohmic contact with the n-contact layer 150. As is well-known in the art, LED structures typically contain doped n- and p-type transport layers, and more heavily doped n- and p-type contact layers. They serve a few different purposes. Forming ohmic contacts to semiconductors is simpler if the semiconductors are doped. Since the emitter layer is typically intrinsic or lightly doped, it is much simpler to make ohmic contacts to the doped transport layers. As a result of surface plasmon effects (K. B. Kahen, Appl. Phys. Lett. 78, 1649 (2001)), having metal layers adjacent to emitter layers results in a loss of emitter efficiency. Consequently, it is advantageous to space the emitter layers from the metal contacts by sufficiently thick (at least 150 nm) transport layers. Next it is advantageous to employ transport layers that not only can easily inject charge into the emitter layer, but also prevent the carriers from leaking back out of the emitter layer. As a consequence, the transport layers will have the largest bandgaps of the device layers. As is well known in the art, highly doping wide bandgap semiconductors is difficult as a result of self-compensation effects. Consequently, forming ohmic contacts to these layers can prove to be difficult. As a result, it is adventitious to add contact layers to the device whose bandgap is smaller than that of the transport layers. Beyond these advantages, doping the transport layers also reduces ohmic heating effects (which can be highly important for laser devices) and leads to larger separations of the corresponding n- or p-Fermi levels (which also aids laser, pin diode, and photovoltaic devices). The above discussion illustrates that having the ability to create doped semiconductor junctions results in numerous advantages for many semiconductor electronic devices.
LED devices have been made since the early 1960s and currently are manufactured for usage in a wide range of consumer and commercial applications. The layers comprising the LEDs are conventionally based on crystalline semiconductor materials that require ultra-high vacuum techniques for their growth, such as, metallo-organic chemical vapor deposition (MOCVD). In addition, the layers typically need to be grown on nearly lattice-matched substrates in order to form defect-free layers. These crystalline-based inorganic LEDs have the advantages of high brightness (due to layers with high conductivities), long lifetimes, good environmental stability, and good external quantum efficiencies. The high conductivities of the transport layers result from high mobilities (due to the crystalline nature of the films) and the ability to readily dope crystalline layers with donors and acceptors. The usage of crystalline semiconductor layers that results in all of these advantages also leads to a number of disadvantages. The dominant ones are high manufacturing costs, difficulty in producing multi-color output from the same chip, and the need for high cost and rigid substrates.
A way for forming low cost LEDs began in the 1980s with the introduction of organic light emitting diodes (OLED) (Tang et al, Appl. Phys. Lett. 51, 913 (1987)). The transport layers for these devices are highly resistive (108 ohm-cm) in comparison with those used in crystalline LEDs. Recent attempts at doping these layers (J. Huang et al., Appl. Phys. Lett. 80, 139 (2002)) have resulted in layer resistivities in the 104-106 ohm-cm range. However, many of these dopants are unstable and the resistivities are many orders of magnitude higher than crystalline LED values of ˜0.1 ohm-cm. The result of employing resistive layers is that one suffers from ohmic heating effects; it is difficult to make ohmic contacts; and since the drive current of the device is limited, so is the overall brightness of the device.
The above examples illustrate that higher performance semiconductor devices can be created from crystalline semiconductor materials; but with the drawback of high manufacturing costs. Attempts to reduce the manufacturing costs by employing organic materials result in lower performance devices whose specifications sometimes fall significantly short of market requirements (e.g., organic-based photovoltaics). Two approaches to lower the cost of crystalline semiconductor materials are to employ either amorphous or polycrystalline inorganic semiconductor materials; however, both of these approaches have well-known drawbacks. Taking the case of devices formed from amorphous Si, both thin-film transistor and photovoltaic (PV) devices have significantly reduced performance due to low mobilities (and the Staebler-Wronski effect for PVs). The performance of polycrystalline-based devices is improved with devices formed from processes, such as, sputtering and CBD. However, sputtering is a higher cost, vacuum-based deposition process and CBD, though chemically based, has long deposition times and is inefficient in its usage of starting materials, as stated previously.
The newest way for creating low cost semiconductor devices is to form the layers from inorganic semiconductor nanoparticles. To obtain the full advantage of these crystalline particles for usage in semiconductor transport layers, the nanoparticles should both be doped (to increase their intrinsic carrier concentration) and devoid of organic ligands on their surface (which impede charge transport). In spite of a plethora of reports about doping colloidal nanoparticles to modify their emission and magnetic characteristics (S. C. Erwin et al., Nature 436, 91 (2005)), there has been very limited research devoted to modifying the nanoparticle's carrier concentration (D. Yu et al., Science 300, 1277 (2003)). In the work of Yu et al. (D. Yu et al., Science 300, 1277 (2003)), even though they doped nanoparticle films, it was done by adding potassium through a high vacuum, post deposition, vacuum evaporation process. In general, even if nanoparticles are stripped of their insulating organic ligands by an annealing process, without added impurities atoms to modify the donor or acceptor concentrations, the resulting nanoparticles have limited conductivities (I. Gur et., Science 310, 462 (2005)). As a result, the corresponding semiconductor junctions incorporating these nanoparticles would have sub-par performance characteristics.